Explore HFT Data | NVIDIA Minerva User Guide

After an HFT collection runs, explore its 10-second summaries and raw samples in Trends. Raw samples are available until the raw-sample retention period ends (see Limits). To explore a collection:

On the High Frequency Telemetry page, select the collection and click Explore. Trends opens on the HFT collection view, filtered to the collection's time range, switches, and interfaces.
The Timeseries tab shows the 10-second summaries. Each counter is a value of the metricName dimension, for example hft_port_rx_bytes for rx-byte. The summary measures include summaryMin, summaryMax, summarySum, summaryCount, and dataPointCount.
Select the HFT tab to see the raw samples. Each counter is a measure here, for example hft_port_rx_bytes. One measure is shown by default; select more to compare them. The time resolution adjusts as you zoom. Use the Navigator below the chart to zoom in and move along the collection. To compare ports, switch priorities, traffic classes, or priority groups side by side, click the dimension to pin it in Comparisons.

Explore builds the HFT collection view from the collection each time you open it. To keep it, save it as your own view. You can use the summaries like other Trends data, including in views that send notifications.

Counter and Metric Names

In Trends, each counter appears under a metric name. Every metric has the deviceName and ifName dimensions. Traffic-class, switch-priority, and priority-group counters also have the trafficClass, switchPriority, or priorityGroup dimension.

Counter	Metric
`rx-byte`	`hft_port_rx_bytes`
`tx-byte`	`hft_port_tx_bytes`
`rx-packet`	`hft_port_rx_packets`
`tx-packet`	`hft_port_tx_packets`
`if-in-discards`	`hft_port_in_discards`
`pause-mac-ctrl-frames-rx`	`hft_port_pause_rx`
`pause-mac-ctrl-frames-tx`	`hft_port_pause_tx`
`ar-grade-high`	`hft_ar_grade_high`
`ar-grade-low`	`hft_ar_grade_low`
`tc-occupancy`	`hft_tc_occupancy`
`tc-watermark`	`hft_tc_watermark`
`tc-ecn-marked`	`hft_tc_ecn_marked`
`tc-octets-tx`	`hft_tc_tx_octets`
`sp-octets-rx`	`hft_prio_rx_octets`
`sp-pause-rx`	`hft_prio_rx_pause`
`sp-pause-tx`	`hft_prio_tx_pause`
`pg-occupancy`	`hft_pg_occupancy`
`pg-watermark`	`hft_pg_watermark`
`pg-headroom-occupancy`	`hft_pg_headroom_occupancy`
`pg-headroom-watermark`	`hft_pg_headroom_watermark`

Read-only users can't open collections, but they can explore HFT data. In Trends, change the data source to the HFT data source.

Last updated: October 29, 2026